Cookieless delivery of personalizied content

This application is a continuation of U.S. application Ser. No. 18/116,872, filed Mar. 3, 2023, and issued as U.S. Pat. No. 11,960,551 on Apr. 16, 2024, which is a continuation of U.S. application Ser. No. 16/885,695, filed May 28, 2020, and issued as U.S. Pat. No. 11,599,585 on Mar. 7, 2023, the disclosures of which are both incorporated by reference in their entirety.

With the advent of privacy restrictions, traditional methods of on-line user data collection and corresponding user-personalized content production are dramatically changing. The traditional way involves gathering data about users, their behavior via cookies, i.e., first party (1P) cookies, (cookies created by the domain a web user is visiting), and third party (3P) cookies, (cookies created by domains other than the one the user is visiting at the time, and are mainly used for tracking and online-advertising purposes), segmenting the audience and then targeting the channels. Web browser providers are embarking on initiatives to stop user data leaking into the backend. This started off with identifying 3P cookies but will soon expand to stopping 1P tracking cookies as well.

Generally, cookies are small text files placed on user devices after they visit a website. The information cookies contain is then accessed by servers on the visited site. The data cookies carry makes it possible to identify and recognize users later. The only threat cookies may pose relates to user privacy, if the cookies are employed for illicit purposes.

Typically, cookies are used for many different reasons, for example: session management for logins, shopping carts, game scores; user privacy controls & settings; user profiling, segmentation, optimization; analytics, attribution, verification; mapping users across platforms; ads frequency capping; and targeting & retargeting. The use of cookies has a large effect on user experience by making web browsing more convenient and personalized.

1P cookies are issued by a website that a user navigates to and views directly. When a user loads a website in a browser, for example, foo.com, then this site creates a cookie which is then saved in the browser data of the user's computer. Third-party (“3P”) cookies are not created by the website being visited, but rather by another entity. For example, if a user accesses a page at foo.com, and the page includes a video hosted at video-hosting site example.com, the video hosting site example.com may set a cookie which is then saved in the browser data of the user's computer.

In such a situation the website owner, foo.com, may embed a piece of code and a video provided by example.com on the relevant foo.com page. When the video hosting site code is executed in the user's browser, or the video is loaded, example.com may track the video player and put data in cookies stored on the user's browser data. The cookie is therefore classified as a third-party cookie being created by a domain other than foo.com.

3P cookies may be used heavily in online advertising where advertisers or web service advertising entities that service retail clients add their 3P tags to a page that may display ads, as well as track users and user devices across different sites users visit.

However, the advent of data privacy features in browsers such as Safari's Intelligent Tracking Prevention (ITP) have incorporated methods to assess which privately controlled domains may track users across different websites. ITP utilizes a machine-learning model, (known as the Machine Learning Classifier), which is fed statistics collected by a users' Safari browser, and when the Machine Learning Classifier identifies that a particular 1P cookie could potentially be used for tracking, the 1P cookie will be blocked. These actions impact the advertiser's ability to provide reporting, affiliate marketing and attribution techniques such that websites may no longer leave cookies in the user's browser for later retargeting and attribution purposes. This impacts companies that access their 1P cookies in a third-party context, and as a result, compromises reporting capability and accuracy.

As previously described, it may be desirable to perform tracking a user's browsing history for many purposes, one of which may be to provide directed or personalized media content, for example, advertising content, to a user in a privacy-compliant manner. Embodiments disclosed herein provides systems and techniques that allow for useful functionality, such as relevant or personalized media content, user preference retention, and the like, which occur within such privacy-compliant contexts. For example, one method presented herein includes sending web-browser executable instructions to a local machine of a user that includes a prioritized query index, the queries in the query index may be run against privacy-protected user data and user-attribute data to determine any matching user data to the search criteria of a query. When a query identifies a match to any of the user data, the browser-executable instructions may request an external personalized content provider or a media content provider to send personalized content or media content, respectively, to the local machine browser based on the matching user data.

illustrates an example of a modelfor tracking multiple user sourcesof on-line user behavior, i.e., typically user-specific data, to generate targeted user-personalized content production provided either directly to the users via a media content provider(s), or through a personalization content providerfor delivery and display to a website, e.g., “SITE1.COM”.

The modelincludes receiving user-specific datafrom various sources, for example, users 1-6,, via tagsat a tag collector, wherein the tagsare generated by users visiting,a website,and, respectively. Typically, the identification of users is provided by either 1P or 3P cookies. This user-specific data is typically collectedinto large data corpus warehouse repositories. The collected user data in the data corpusis used by marketing entities, e.g., marketer Aand marketer B, who segment and identify users from the collected user data in the data corpusto producespecific user attribute-based audiences.

These attribute-based audiences are provided to a personalization content providerto provide personalized contentto websites, e.g., “SITE1.COM”, to be displayed when a user from the user sourcesis determined to be a member of a particular audience. The attribute-based audiences are further provided to media content providersto provide targeted mediato users from the user sourceswho are determined to be a member of a particular audience, for example in, User1 and User 4.

However, the technique ofmay not be viable, or may be undesirably inefficient, given the shift in the web browser provider industry by adopting the ITP practices and protocols.

Web browser providers to limit user data stored on a user's local machine browser in their browsers. Thus, content providers may not be able to track or receive user behavior outside of the user's browsers. Web browser providers may instead allow browser-executable code to be executed in their browsers to run routines to determine if a user matches a certain set of well-defined behaviors, e.g., the gender of the user, or the user visits a particular website, etc.

In the data processing world, a traditional way of processing data may be to bring data to tables, curate and index the tables, and run queries against the data in the table. When there are many queries, the data may be optimized, and each query may be run on top of the optimized data to get results for each query. This works well when all the data is accounted for and when new queries are received.

In situations when data is received, for example, in an event, but there is no large data corpus available to optimize, that is, when there is a known query but the data may not be present, Query Indexing/Streaming provides a solution by assembling all the known queries into a query model, so that when a new data arrives, a query may be quickly determined to match the incoming data based on the query model and rules may be fired accordingly.

A representative solution provided herein instead pushes a query to a client. A Data Management Platform (DMP)/Customer Data Platform (CDP) system may no longer merely collect data, but instead function as a query indexing engine. CDPs typically may be used for creating personalized customer experiences by collecting and tying together customer data through personally identifiable information (PII)—like email addresses and phone numbers—to create a 360-degree view of the customer. The primary data source for CDPs is first-party data from customers who have directly interacted with the business online (through website interactions, campaign engagement, online purchases, and loyalty programs) as well as offline (through in-store purchases, in-person events). CDPs may also be able to use second-party data (sourced from businesses that collect and sell first-party data) and third-party data (collected through anonymous identifiers like cookies) in addition to first-party data. DMPs typically collect primarily anonymous data to profile, analyze, and target online customers; these platforms help digital marketers make more informed media buying decisions and more effectively target campaigns. Digital marketing agencies and in-house marketing teams use DMPs to identify audiences by categories like demographic, behavior, or location to better target digital media content campaigns. DMPs aggregate high volumes of anonymous customer data originating from multiple sources. The primary data sources for DMPs are second- and third-party data. DMPs must work with anonymous entities like cookies, devices, and IP addresses to exchange information about audiences while protecting personal privacy.

In, for example, media content providers, represented by marketer Aand marketer B, may generate segment-based querieswith respect to user data stored in the large data corpus, where the segment-based queries are directed toward particular market segments of users the marketers,intend to provided targeted media content. For example, media content providers may create the following queries: a.) target all male customers, living in Seattle, WA, greater than 40 years of age; and b.) target all female customers, living in San Francisco, CA, who frequently visit on-line shopping sites.

A DMP/CDP systemmay assemble each query into a prioritized decision tree graph query index that may then be constructed into a browser-executable library that may be sent to any browsersvia websites, for exampleand a personalization content provider. The prioritized decision tree graph queries may include libraries of queries, or sub-collections of queries pushed down to local browsers. Alternatively, the browser-executable library may be constructed by compiling the query index with other browser-executable instructions into the browser-executable library.

The assembled query indices may be stored in a query indices repositoryfor subsequent transmission to user local machine browsers. A query resolver executing by means of the received browser-executable library in the local machine browser may then validate which query(ies) in the executable library query index match user data stored in the user's web browser. When a match or query hitis detected, the browser fires an appropriate call to either the personalize content providerto media content providersto send corresponding content to the content requesting browser.

Thus, the media content marketers,prioritize market segment queries based upon a data set from the data corpusby the DMP/CDP systemperforming query optimization of a prioritized market segment set of queries to create a query index. The entire set, or a subset of the decision tree graph query (based on size), is sent to a user browser via the personalization content provideror via a website (e.g., “SITE2.com”).

When the user's browserloads the website, for example, SITE2.com, the browser-executable library in the webpage uses the decision tree graph query to determine which of the executable library queries may match the user's behavior stored in the browser as user-specific data. When a query hitis determined, the browsermay call back to the personalization content providerto retrieve for display in the user's browserappropriate personalized content. In the alternative, when a query hitis determined, the browsermay in addition make a call to a media content providerwhich may retrieve and send an appropriate media content for display on the user's browser.

The methods presented herein may be directed toward pushing to a local user's browser: 1) a browser-executable library that causes the browser to locally process a query transmitted with the browser-executable library and extract data from the user's browser related to user attribute data and the user's browser activity, both in a current browser session and previous browser sessions, wherein the browser-executable library causes the user-extracted data to be sent to a remote server; and pushing to a local user's browser, 2) a query that provides instructions to the browser-executable library running in the local browser to collect particular types of user data related to user attributes and historical user activities in previous and current sessions of the browser.

Another feature of the methods presented herein may be the shifting the processing of cookie-based data collection and subsequent content distribution at remote servers to a client's browser. Prioritized decision tree graph queries are pushed to user's browsers within browser-executable libraries that when executed, extract user data specifically related to the matching queries from the transmitted prioritized decision tree graph queries.

User data processed by the browser-executable library running the user's browser may also provide derived data associated with particular categories of predefined user behavior, for example, customers who like motorcycles, customers who like red shoes, etc.

illustrates workflow diagramof the methods presented herein. A local machine browserexecuting on a user's local computer may include a browser User Interface (UI), an DMP/CDP libraryfor retrieving and executing browser-executable library containing prioritized decision tree graph queries, an DMP/CDP query repositoryfor receiving and storing the prioritized decision tree graph queries, and user-specific browser datastored in relation to a user's historical and current browser sessions. The workflow diagramfurther includes a remote DMP/CDP server(s), (similar to DMP/CDP systemof), that provides the prioritized decision tree query in the browser-executable library to the user local machine browser, and includes an external content serverthat may receive a request for and dispense external content to the user location machine browser.

The workflow diagrammay begin with a user on navigating within the browser UIof the user local machine browserto a particular Uniform Resource Locator (URL) to begin loadinga webpage associated with the URL into the browser UI.

The browser UIqueries the DMP/CDP libraryto retrieve an execution code library specific to the URL requested by the browser UI. If the execution code library exists in the library, the execution code library begins runningwith the hyper-text markup language (HTML) of the requested URL. If the execution code library specific to the URL does not exist at the library, the execution code library may alternatively be requestedby the libraryto the DMP/CDP server(s)and accordingly returnedto the libraryto begin runningwith the HTML of the URL.

Under control of the execution code library, the DMP/CDP librarythen readsthe Document Object Model (DOM) of the returned from the URL to determine how to parse data attributes of the HTML data returned from the URL, and performs a privacy checkwith the browser UI. If the privacy check determines a “don't track” attribute is present, the attribute is presentedto the librarywhich then may restrict certain data tracking functions.

Under control of the execution code library, the DMP/CDP libraryfurther requestsa prioritized decision tree query to be transmittedfrom the DMP/CDP queries repositoryin the user local machine browserback to the library. However, if an appropriate prioritized decision tree query is not found at the query repository, the librarymay requestanother prioritized decision tree query to be transmittedfrom the remote DMP/CDP server(s)back to the libraryat the user local machine browser.

The queries returned to the DMP/CDP librarymay be matched against user attribute data to produce a prioritized set of queries based on the user attribute data. An example query may include, for example, but not limited to, a user who is interested in sports, a user who has seen a particular media content, or a user who has navigated to a particular website.

Under control of the execution code library, the DMP/CDP libraryfurther utilizes a user local machine browserApplication Programming Interface (API) to establish a privacy sandbox to applythe queries retrieved from the query repositoryor the remote serveragainst user-specific browser data stored in a browser date repositoryindicating historical data of the user's interaction with the user local machine browserthrough the browser UI.

A browser “privacy” sandbox introduces a set of privacy-preserving APIs to accomplish tasks that are typically used in tracking user data. These security APIs enable the user's browser to act on the user's behalf to ensure that data is never shared without their knowledge and consent. The security APIs enable use cases such as ad targeting and conversion measurement, but without revealing individual private and personal information.

User-specific browser data is returnedto the librarythat is associated with all user-specific data matching the applied queries from which audience segments are determinedby the executing code library running on the DMP/CDP libraryof the user local machine browser. The audience segment information is addedto external tags in the HTML of the requested URL in the browser UI

The browser UIthen requestscontent or media content by serving URLs with the audience segment tag information from external server(s)which then returnsthe requested content or media content to the browser UIof the user location machine browser.

The browser UIthen finishesparsing the DOM and comes to a “DOM ready” event. Thereafter, the browser UIfinishes loadingany remaining content of the requested URL and comes to a “Page Load” event after the URL page is fully load in the browser UI.

Under control of the execution code library, the DMP/CDP libraryfurther transmits, to the DMP/CDP server(s), user level identifiers capable of parsing for web-analytics, and URL target dependent data identifying particular websites that have been visited. The libraryfurther transmits, to the server(s), a request to returnupdated queries to the query repositoryat the user local machine browser.

Under control of the execution code library, the DMP/CDP library, further transmitsany data stored in the browser to the browser UI, that may include, a user ID, a number of visits to a URL, a last visit to a URL, an audience segment of the user, etc.

illustrates a logic diagram of an example of query data flowof another aspect of the methods presented herein.

The query flow diagram begins with a user requestinga target site homepage, for example, by typing the URL “http://foo.com” into a browser. An execution code library specific to the “foo.com” URL requested by the browser is executedat the browser that looks at user-specific data to determine certain user-attributes relevant to queries executed against the user-specific data. The example shows a query servicebeing performed by the execution code library operating at the DMP/CDP library, (as shown in), that returnstwo queriesandbased on two corresponding user attributes detected in the browser dataof the user local machine browser. For example, a user attribute of a “return visitor”may correspond to query A, and another user attribute of an “email subscriber”may correspond to query B.

The flow diagram then proceeds to run query Aagainst user specific data in a privacy sandbox of the browser data, (as shown in), to determineif a specific media content was viewed by the user in the browser UI. If it is determined that the specific media content was indeed viewed, then the system retrieves for displayon the browser UIcontent corresponding to a personalized media treatment A.

If it is determined that the specific media content was not viewed by the user in the browser UI, then flow diagram proceeds to run query Bagainst user specific attributes (for example, an attribute indicating if the user is an email subscriber to a particular interest group), to determineif the user is in a specific interest group. If it is determined that the user is in the specific interest group, then the system retrieves for displayon the browser UIcontent corresponding to a personalized media treatment B. In the alternative, if it is determined that the user is not in the specific interest group, then the system retrieves for displayon the browser UIcontent corresponding to a personalized media treatment C.

illustrates a logic flow diagram of a methoddisclosed herein. The computer-implemented method includes generatinga query index from a data corpus, the query index including a plurality of market segment-based queries, wherein each market segment-based query of the plurality of queries is configured to provide targeted content on a browser user interface of a user determined to be within a corresponding market segment.

The method further includes constructinga browser-executable library including the query index, wherein the browser-executable library is configured to execute within a local machine browser of the user, and then transmittingthe browser-executable library to the local machine browser of the user, wherein the browser-executable library is configured to determine that a query of the plurality of market segment-based queries matches user-specific data stored in the local machine browser of the user.

The method further includes causingthe local machine browser to request the targeted content corresponding to the user-specific data based on the query matching the user-specific data stored in the local machine browser of the user.

The method further includes the user-specific data stored in the local machine browser including user browser history data stored in a privacy sandbox of a browser data repository, wherein the user browser history data indicates a behavior of the user interacting with the local machine browser.

The method further includes the user-specific data stored in the local machine browser including user-specific attribute data.

The method further includes the requested and received targeted content containing user-specific personalized content.

The method further includes the requested and received targeted content containing user-specific content.

The method further includes replying to a request to transmit an updated query index from the local machine browser.